Google Trends

Loading Packages

library(tidyverse)

## Warning: replacing previous import 'vctrs::data_frame' by 'tibble::data_frame'
## when loading 'dplyr'

## ── Attaching packages ──────────────────────────────────────────────────────────────────────────────────────────────── tidyverse 1.3.0 ──

## ✓ ggplot2 3.3.2     ✓ purrr   0.3.4
## ✓ tibble  3.0.3     ✓ dplyr   1.0.1
## ✓ tidyr   1.1.1     ✓ stringr 1.4.0
## ✓ readr   1.3.1     ✓ forcats 0.5.0

## ── Conflicts ─────────────────────────────────────────────────────────────────────────────────────────────────── tidyverse_conflicts() ──
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

library(plotly)

## 
## Attaching package: 'plotly'

## The following object is masked from 'package:ggplot2':
## 
##     last_plot

## The following object is masked from 'package:stats':
## 
##     filter

## The following object is masked from 'package:graphics':
## 
##     layout

library(gtrendsR)

Loading and cleaning Google Trends data

## Name order data
name_order = read.csv("./data/name_order.csv", skip = 2, col.names = c("state", "biden_trump", "trump_biden")) %>% 
  mutate_all(funs(str_replace(., "%", ""))) %>% 
  mutate(
    biden_trump = as.numeric(biden_trump),
    trump_biden = as.numeric(trump_biden)
  )

## Warning: `funs()` is deprecated as of dplyr 0.8.0.
## Please use a list of either functions or lambdas: 
## 
##   # Simple named list: 
##   list(mean = mean, median = median)
## 
##   # Auto named with `tibble::lst()`: 
##   tibble::lst(mean, median)
## 
##   # Using lambdas
##   list(~ mean(., trim = .2), ~ median(., na.rm = TRUE))
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_warnings()` to see where this warning was generated.

## Today's searches for "voting"
#voting = read.csv("./data/voting.csv", skip = 2, col.names = c("state", "vote_voting"))

voting = read.csv("./data/voting_121920.csv", skip = 2, col.names = c("state", "vote_voting"))

## Change in voting searches since last election

voting_percentages = read.csv("./data/state_names.csv") 

voting_percentages$state_code = paste0('US-', voting_percentages$state_abbreviation)

voting_percentages = 
  voting_percentages %>% 
  mutate(
    data = map(state_code, ~(gtrends(keyword = "vote + voting", geo = .x, time = '2016-09-30 2020-10-31')$interest_over_time))
  )

for (i in 1:51) {
    voting_percentages$data[[i]]$hits = str_replace(voting_percentages$data[[i]]$hits, "<1", "0")
    voting_percentages$data[[i]]$hits = as.numeric(voting_percentages$data[[i]]$hits)
}

voting_percentages =
  voting_percentages %>% 
  unnest(cols = data) %>% 
  select(state, state_abbreviation, date, hits) %>% 
  mutate(date = as.Date(date)) %>% 
  filter(
    date <= "2016-10-29" | date >= "2020-10-01" ## Change to october 4th
    )

voting_percentages = 
  voting_percentages %>% 
  separate(date, c("year", "month", "day"), "-")

voting_percentages = 
  voting_percentages %>% 
  group_by(state_abbreviation, year) %>% 
  summarize(october_average = mean(hits)) %>% 
  pivot_wider(names_from = year,
              values_from = october_average)

## `summarise()` regrouping output by 'state_abbreviation' (override with `.groups` argument)

voting_percentages = 
  voting_percentages %>% 
  rename(
    october_2016 = "2016",
    october_2020 = "2020"
  ) %>% 
  mutate(
    percent_change = ((october_2020 - october_2016)/october_2016)*100
  ) %>% 
  mutate(
    percent_change = round(percent_change)
  )

Map 1: Name order

name_order$hover = with(name_order, 
                        paste(state, '<br>',
                              '"Trump Biden" =', trump_biden,"%", '<br>', 
                              '"Biden Trump" =', biden_trump,"%"))

g = list(
  scope = 'usa',
  projection = list(type = 'albers usa'),
  showlakes = FALSE,
  lake_color = toRGB('white')
)

map1 = plot_geo(name_order, locationmode = "USA-states", width=800, height=400)

map1 = 
  map1 %>% 
  add_trace(
    z = ~trump_biden, text = ~hover, locations = ~state, 
    color = ~trump_biden, colors = 'Reds',
    hoverinfo = "text"
  )

map1 = map1 %>% colorbar(title = '%"Trump Biden"')

## Warning: `arrange_()` is deprecated as of dplyr 0.7.0.
## Please use `arrange()` instead.
## See vignette('programming') for more help
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_warnings()` to see where this warning was generated.

map1 = map1 %>% layout(
    title = "Order of Presidential Candidates' Names in Google Searches<br>(Hover for breakdown)",
    geo = g
  )

map1

Map 2: Voting Searches

voting$hover = with(voting, 
                        paste(state, '<br>',
                              'Relative Proportion =', vote_voting))

map2 = plot_geo(voting, locationmode = "USA-states", width=800, height=400)

map2 = 
  map2 %>% 
  add_trace(
    z = ~vote_voting, text = ~hover, locations = ~state, 
    color = ~vote_voting, colors = 'Blues',
    hoverinfo = "text"
  )

map2 = map2 %>% colorbar(title = 'Relative Search Vol.')

map2 = map2 %>% layout(
    title = 'Google Searches Containing "vote" or "voting"',
    geo = g
  )

map2

Map 3: % Change in Voting Searches

voting_percentages$hover = with(voting_percentages, 
                        paste(state_abbreviation, '<br>',
                              '% Change =', percent_change,'%'))

map3 = plot_geo(voting_percentages, locationmode = "USA-states", width=800, height=400)

map3 = 
  map3 %>% 
  add_trace(
    z = ~percent_change, text = ~hover, locations = ~state_abbreviation, 
    autocolorscale = TRUE,
    hoverinfo = "text"
  )

map3 = map3 %>% colorbar(title = '% Change')

map3 = map3 %>% layout(
    title = 'Percent Change in Google Searches Containing "vote" or "voting"<br>October 2020 vs. October 2016',
    geo = g
  )

map3

#Sys.setenv('plotly_username'='xxxx')
#Sys.setenv('plotly_api_key'='xxxx')
#api_create(map1, filename = 'map1', sharing = 'public')
#api_create(map2, filename = 'map2v2', sharing = 'public')
#api_create(map3, filename = 'map3v2', sharing = 'public')

Correlation calculation

correlation = read_csv("./data/correlation.csv")

## Parsed with column specification:
## cols(
##   state_abbreviation = col_character(),
##   percent_change = col_double(),
##   turnout_change = col_double()
## )

correlation =
  correlation %>% 
  mutate(
    percent_change = log(percent_change),
    turnout_change = log(turnout_change)
  )

cor.test(correlation$percent_change, correlation$turnout_change, method = "pearson") 

## 
##  Pearson's product-moment correlation
## 
## data:  correlation$percent_change and correlation$turnout_change
## t = -1.2099, df = 49, p-value = 0.2321
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.4259150  0.1104447
## sample estimates:
##        cor 
## -0.1703231

correlation %>% 
  ggplot() +
  geom_point(aes(x = percent_change, y = turnout_change)) +
  labs(
    title = "% Change in Google searches vs. % Change in voter turnout",
    x = "% Change in Google searches, 2016-2020",
    y = "% Change in voter turnout, 2016-2020"
  ) + 
  theme_minimal()

0.1703231*0.1703231

## [1] 0.02900996